Skip to main content

Full text of "Improvements in Raman Lidar Measurements Using New Interference Filter Technology"

See other formats


Source of Acquisition 
NASA Goddard Space Flight Center 



Improvements in Raman Lidar 
Measurements Using New Interference 

Filter Technology 

Dr. David N. Whiteman 

NASA/GSFC, Greenbelt, MD 20771 

david.n.whiteman@nasa.gov 

Mr. John R. Potter 

Barr Associates, Westford, MA 01 886 

jpotter@barrassociates.com 

Ms. Rebecca Tola 

Barr Associates, Westford, MA 01886 

rtola@barrassociates.com 

Dr. Igor Veselovskii 

University of Maryland, Baltimore County, Baltimore, MD 21250 

iveselov@agnes.gsfc .nasa. gov 

Mr. Martin Cadirola 

Ecotronics, LLC, Clarksburg, MD 20871 

cadirola@ecotronics.com 



Mr. Kurt Rush 

NASA/GSFC, Greenbelt, MD 20771 

kurt.d.rush@nasa.gov 

Mr. Joseph Comer 

Science Systems and Applications Inc, Lanham, MD 2070 

comer@agnes.gsfc.nasa.gov 



1 Abstract 



Narrow-band interference niters with improved transmission in the ultra-violet have been devel- 
oped under NASA- funded research and used in the Raman Airborne Spectroscopic Lidar (RASL) 
in ground-based, upward-looking tests. Measurements were made of atmospheric water vapor, 
cirrus cloud optical properties and carbon dioxide that improve upon any previously demonstrated 
using Raman lidar. Daytime boundary and mixed layer profiling of water vapor mixing ratio up to 
an altitude of approximately 4 km is performed with less than 5% random error using temporal and 
spatial resolution of 2-minutes and 60-210, respectively. Daytime cirrus cloud optical depth and 
extinction-to-backscatter ratio measurements are made using 1 -minute average. Sufficient signal 
strength is demonstrated to permit the simultaneous profiling of carbon dioxide and water vapor 
mixing ratio into the free troposphere during the nighttime. A description of the filter technology 
developments is provided followed by examples of the improved Raman lidar measurements. 



2 Introduction 

Raman Lidar is now regarded as one of the leading technologies for atmospheric profiling of water 
vapor [Melfi et. al, 1989] [Whitemaii et. al., 1992] [Turner et. al., 2000], cirrus clouds [Ansmann et. al., 1992j 
[Reichardt et. al, 2002] [Whitemanet. al, 2001a] [Whitemanet. al., 2004] and other quantities 
including aerosols [Ansmann et. al., 1990] [Ferrare et. al., 2005] and temperature [Arshinov et. al., 2005] 
[Behrendt et. al., 2002] [Di Girolamo et. al, 2004]. Experimental measurements using Raman li- 
dar have been made of carbon dioxide [Riebesell, 1990] [Ansmann et. al, 1992b] as well. Tradi- 
tionally, most Raman lidar measurements based on laser sources in the near-UV (approximately 
350 nm) were limited to the nighttime. In the 1990s, advances in Raman lidar technology (high- 
power UV lasers and narrow-band interference filters) and techniques (narrow field-of-view detec- 
tion) resulted in systems operating in the near-UV that measure water vapor and aerosols through- 
out the diurnal cycle [Turner et. al., 2000]. 

Interference filtering was one of the crucial technologies that permitted this extension of Ra- 
man lidar water vapor measurements to the daytime. Prior to the research undertaken here, day- 
time measurements of water vapor mixing ratio were typically made using interference filters with 
widths of 0.25 - 0.3 nm and peak transmissions of 30-40% at a wavelength of -407.5 nm. It was 
therefore clear that significant lidar system performance gains could be realized through improve- 
ment in peak transmission of the water vapor interference filter being used since the statistical 



uncertainty in the water vapor measurement dominates the error budget of a Raman Lidar water 
vapor mixing ratio measurement. Other measurement techniques, both passive and active, could 
also benefit from higher transmission interference filters due to the widespread use of interference 
filter technology in optical detection systems. For these reasons, the National Aeronautics Space 
Administration's (NASA) Advanced Component Technology (ACT) program funded a joint re- 
search effort between Barr Associates and NASA/Goddard Space Flight Center to develop and test 
advanced narrow-band, UV interference filters with up to twice the peak transmission while main- 
taining all other specifications of the filter the same. This implied that the physical size and weight 
of the filters and their rejection of out-of-band light would not change finr previous technology. 
Therefore, the goal of the research was to develop and validate interference filters that significantly 
improve the measurement capability of remote sensing systems including lidars but do not add to 
the weight, volume or power requirements of the system. 

3 Interference Filter Technology Development 

The objectives of this research were to develop a process for fabricating UV filters with significant 
improvement in throughput, then to build and demonstrate a series of UV filters possessing 100 
to 250 picometer bandwidths and peak transmission of 60-80%. Temperature stability and long 
lifetime were desired indicating that refractory oxides would be the thin film materials of choice. 
Among the filters developed under this activity were ones to measure Raman scattering from water 



vapor, nitrogen and carbon dioxide when excited by the frequency-tripled Nd:YAG laser (354.71 
nm). The specifications of these 50 mm diameter filters are shown in Table 1. 

Various materials that have previously been found useful for fabrication of UV interference fil- 
ters [Schink et. al., 1995] [MacLeod et. al, 1989], including tantalum pentoxide (Ta 2 05), hafnium 
oxide (Hf0 2 ) and zirconium oxide (ZrC>2), were studied as a part of this effort. Depositions were 
made using each of these materials in combination with silicon dioxide and then spectral com- 
parisons were made. Three deposition processes were explored in order to make thin films using 
these materials. The focus was on optimizing these processes for minimal absorption, scattering 
and reflection losses on Fabry-Perot narrowband interference coatings. Electron beam, Kaufman 
ion- assisted deposition (IAD) [Pawlewicz, 1998], DC magnetron sputtering and ion beam sputter- 
ing (IBS) processes were compared as methods to deposit thin films for this research effort. Of the 
three processes investigated, the Kaufman IAD process yielded results closest to the design goals. 
Scattering losses on both the ion beam and magnetron sputter processes proved to be too great to 
achieve high throughput on ultra-narrowband UV filters. It was shown by initial experimentation 
that Zirconium dioxide (Zr0 2 ) had excessive scattering losses, as compared with Ta 2 5 . The major 
effort was, therefore, expended in optimizing filters constructed by layering Ta 2 Os using the IAD 
process. This task was broken down into four separate elements; improving spatial uniformity, 
minimizing scatter losses, minimizing reflection losses and minimizing absorption losses. Each of 
these elements will now be addressed. 



1. Wavelength and bandshape spatial uniformity were addressed with the goal of achieving less 
than 0.005% change in either parameter across a 50 mm diameter substrate. The experimen- 
tation led to three conclusions: a consistent substrate temperature is required during the depo- 
sition process, a substrate significantly larger than the final size of the part greatly improves 
the heat distribution and large clearances in front of and behind the substrate minimized the 
thermal effects on optical thickness distribution. Even at the end of this activity, however, 
consistent fabrication of spatially uniform 0.1 nm wide filters remained a challenge. There- 
fore, further work in process control is necessary to improve the yield of filters possessing 
spatially uniform wavelength and bandshape. 

2. Minimizing scatter is critical to achieving maximum filter throughput. This is particularly 
important with very narrow bandwidth filters operating in the ultraviolet region because scat- 
tering sites are large compared with the filter center wavelength. Coatings were studied with 
a scatterometer demonstrating a correlation between increased scattering and transmission 
loss. As the bandwidth decreased, the transmittance decreased exponentially as a function of 
scattering. Experiments were performed during the manufacture of a 0.1 nm wide filter. The 
surface flatness that was required for maximum transmittance was ~0. 1 nm (r.m.s.), as shown 
in Table 2. If films grew into a crystalline structure, the scattering would increase, therefore it 
was mandatory that the films be amorphous. Substrate temperature, ion beam flux, deposition 
rates and choice of materials all played a role in minimizing scatter. Minimizing contamina- 



tion reduces scattering, so extra precautions were taken to ensure the substrate surface and 
the deposition system did not introduce additional scattering due to contaminants. 

3. Minimizing reflection losses starts with a thin film design where each filter component is 
matched to its media such that the reflectance from each surface approaches zero. Errors in 
thin film thickness as small as 0.005% can increase the reflectance or reduce the transmittance 
by 10 percent. The thickness of the layers was therefore monitored during the fabrication 
process using a collimated halogen light source, blocking filters, UV photomultiplier tube 
and a 1 -meter spectrometer. 

4. Absorption must be minimized in order to obtain maximum transmittance in narrow-band 
interference filters. This is particularly true for layers located near the highest electric field 
in a thin film stack. Process improvements were made to reduce the absorption coefficient of 
tantalum pentoxide (Ta 2 Os) from 1.6 x 10~ 4 to 2.0 x 10~ 5 . Efforts were also concentrated 
on optimizing the ion gun and electron gun parameters along with the partial pressure of 
oxygen during deposition. It was also imperative that the partial pressure of water vapor be 
minimized. 

4 Atmospheric Measurements 

Using the processes just described, filters with specifications shown in Table 1 were manufactured 
at Barr Associates and tested in the NASA/GSFC Mesoscale Atmospheric Processes Raman Lidar 

7 



Laboratory. The filters were installed in the Raman Airborne Spectroscopic Lidar (RASL) which 
was undergoing upward-looking, laboratory-based tests funded through the NASA Instrument In- 
cubator Program. RASL consists of a frequency-tripled Continuum 9050 Nd:YAG laser (17. 5W @ 
354.7 nm), 0.6 m Dahl-Kirkham telescope operated with 0.25 milliradian field-of-view and wave- 
length selection using beamsplitters and interference filters. (More system details on RASL are 

available at http://ramanlidar.gsfc.nasa.gov.) 

4.1 Water vapor mixing ratio measurements 

Water vapor is one of the most important components of the atmosphere from considerations of 
both weather and climate yet it is one of the most difficult to quantify due to its high variability 
on short time and space scales. Advances in water vapor profiling capabilities are sought to im- 
prove quantitative precipitation forecasting [Weckwerth et. al., 2004] and to improve our ability to 
quantify and study mesoscale meteorological systems [Demoz et. al., 2005] [Demoz et. al., 2006] 
[Wulfmeyer et. al., 2006]. Raman lidar is a well-established technology for profiling water vapor 
and other quantities in the troposphere. The technology advances in interference filters achieved 
in this research permits significant improvements in that profiling capability. 

On July 26, 2005 RASL was operated from the ground over a period of approximately 14 
hours from early morning until late in the evening in order to test its upward-looking measurement 
capability. The measurements included periods of wbright mid-summer, daytime conditions in the 
vicinity of Washington, DC where urban haze can significantly increase the sky brightness in the 

8 



visible and near-UV, thus degrading daytime measurement performance. The time series of RASL 
measurements of water vapor mixing ratio, made using filters 1 and 2 listed in Table 1 are shown 
in figure 1. The data were acquired using 1 -minute temporal resolution and 7.5 meter spatial 
resolution. The data were then processed using a 3-minute sliding window in the time domain 
and a sliding window in the vertical domain that varied from 90 to 330 meters. The resulting 
temporal and spatial resolution of the water vapor mixing ratio measurements, determined by the 
half-power point in a Fourier spectral analysis, was approximately 2 minutes and between 60 and 
210 meters, respectively. The measurements were calibrated against the total precipitable water 
measured by a collocated SuomiNet GPS system [Whiteman et. al., 2006b]. High noon occurred 
at approximately 1800 UT when the solar zenith angle reached approximately 20 degrees. The 
daytime boundary layer can be observed in the image at heights that range between 1.5 and 2 km. 
The residual layer, from boundary layer mixing on previous days, is observed to descend from 
approximately 4.5 km to less than 3 km over the period of the measurements. Despite the bright 
conditions present, a moist layer can also be observed descending from 6 km to approximately 5 
km during the measurements. Boundary layer convective cells, which supported the development 
of cumulus clouds at altitudes of 1.5 to 1.8 km, can be observed as vertical striping in the water 
vapor field between 1800 and 2100 UT. The vertical striping of the image at approximately 16.0 
and 19.5 UT are due to clouds that developed at the top of the boundary layer. Times shown with 
values larger than 24 UT are on July 27 (UT). 



A comparison of these RASL measurements made in Greenbelt, MD and a Vaisala RS-80H 
radiosonde launched from the Howard University Research Campus in Beltsville, MD - a distance 
of approximately 10 km from GSFC - is shown in figure 2. The location of features in the vertical 
and the overall calibration of the two measurements are in good agreement. These measurements 
occurred at 1300 UT when the sun was ~20 degrees above the horizon and daytime mixing in the 
boundary layer had not yet developed to a significant degree. Therefore, the water vapor field was 
likely to be reasonably homogeneous between the two sites due to the stable atmospheric condi- 
tions of the previous evening. The radiosonde/lidar comparison shown supports the conclusion that 
the layered features observed in figure 1 are realistic. Furthermore, both figure 2 and comparisons 
of water vapor mixing ratio measurements derived from the first 30-meters of RASL data and those 
from a Paroscientific Met3A sensor (not shown) mounted 10 meters above the laboratory in which 
RASL was located showed good agreement in the lowest portions of the profile even though no 
overlap correction was applied to RASL data. These facts imply that the lidar system overlap func- 
tion, which can require height-dependent corrections to compensate for [Whiteman et. al., 2006a], 
has minimal effect on the RASL mixing ratio measurements presented here. Also, a comparison of 
the time series of total precipitable water vapor measurements from RASL and GPS showed good 
agreement except in the presence of clouds, which attenuated the laser beam and prevented full 
profiling of the atmospheric column. 

The random error in the water vapor mixing ratio data was quantified at three times in figure 

10 



1 to study the evolution of random errors as a function of sun angle and therefore sky brightness. 
Figure 3 presents the RASL water vapor mixing ratio profiles and the random error at 13, 18, and 
26.5 UT when the solar zenith angles were 70, 20, and -12 degrees respectively. The latter value 
indicates that the sun was 12 degrees below the horizon. These profiles possess the same temporal 
and spatial resolution as shown in the image of figure 1 . A general characteristic of the upward- 
looking RASL measurements is the increase in random error below approximately 0.6 km. This is 
due to reduction of the signal in the near field due to the use of a narrow field-of-view detection 
scheme. This is one of the consequences of the single field-of-view design of an airborne lidar 
system intended for downward-looking measurements. A supplemental smaller telescope can be 
used at wider field of view to reduce the near field random errors [Whiteman et. al, 2006a]. 

The profiles of mixing ratio shown on the left side of the figure indicate that on this day the 
boundary layer extended to an altitude of approximately 2 km and was characterized by mixing 
ratio values ranging roughly from 5-15 g/Kg. A significant residual layer existed between altitudes 
of approximately 2 and 4 km where mixing ratio values ranged between ~1 and 7 g/Kg. Above 
the residual layer and up to an altitude of 8 km, mixing ratio values ranged between 1 and 3 g/Kg. 
The random errors are shown on the right side of the figure indicating that, even at 1800 UT (solar 
noon), the random error did not exceed 2% in the boundary layer (except for the near-field zone at 
altitudes less than 0.6 km), 4% in the residual layer and ranged between 20% and 60% above the 
residual layer up to an altitude of 8 km. The measurements acquired at 1300 UT when the sun was 

11 



20 degrees above the horizon possessed less than 3% random error through the residual layer and 
less than 8% below 6 km. The profile acquired at night possessed less than 7% random error up to 
an altitude of 8 km. 

This measurement quality is to be contrasted with an earlier comparison of the relative error 
budgets of differential absorption lidar (DIAL) and Raman lidar water vapor measurements. That 
study [Bosenberg, 2005], based on measurements acquired in Oklahoma in late 1999 by the Max 
Planck Institute (MPI) water vapor DIAL and the U. S. Department of Energy (DOE) Climate and 
Radiation Facility Raman Lidar (CARL), indicated that the Raman lidar was not able to provide 
sufficient temporal and spatial resolution measurements of water vapor in the daytime to permit 
boundary layer turbulence to be studied. It should be noted that at the time of these comparisons the 
CARL instrument used a xlO attenuating filter in the water vapor channel during the daytime due 
to countrate limitations in the data acquisition electronics. The DIAL water vapor measurements 
acquired during this study possessed 3-7% error between the altitudes of 400 and 1500 meters 
using 1 -minute temporal averaging and 90-meter spatial resolution and were of sufficient quality 
to permit turbulence studies to be performed. Extrapolations based on the RASL measurements 
shown in figure 3 indicate that, with the same temporal and spatial resolution and for similar 
water vapor measurement conditions as for the MPI DIAL/CARL study, the RASL random error 
would remain below 4% for altitudes less than 2 km and below 7% for altitudes less than 3km, 
again excluding the near range of ~0.6 km which RASL was not designed to measure with high 

12 



precision. The CARL lidar has been upgraded since 1999 so that it no longer uses an attenuation 
of a factor of 10 in the daytime. CARL daytime water vapor measurements should now be similar 
to those of RASL shown here. Therefore, CARL can now be expected to provide long-term water 
vapor measurements that are suitable for boundary layer turbulence studies. 

These water vapor random error characteristics are improved over recently-published daytime 
Raman lidar water vapor measurements acquired during the International H 2 Experiment (IHOP) 
[Weckwerth et. al., 2004] by the NASA/GSFC Scanning Raman Lidar (SRL) [Whiteman et. al., 2006a] 
[Whiteman et. al., 2006b] . The analysis of errors from that experiment indicated that, under similar 
water vapor and sky brightness conditions, the SRL random error did not exceed 10% throughout 
the boundary layer although the boundary layer rose to greater altitudes. Accounting for the differ- 
ences in the measurements between the SRL during IHOP and RASL on July 26, 2005, the RASL 
measurements exhibit approximately a factor of 2 reduction in the random error in the water vapor 
mixing ratio measurements compared with the SRL during IHOP. This factor of 2 (which requires 
a factor of 4 more signal, for example, if the background skylight remains constant) can be ex- 
plained by the combination of a RASL laser that is approximately twice as powerful as the SRL 
laser and by the newly-developed RASL interference filters that possess approximately twice the 

peak transmission as the filters in use in the SRL at the time of IHOP. 

4.2 Cirrus cloud optical depth and extinction to backscatter ratio 

Cirrus clouds strongly influence the radiation balance of the Earth. Some studies have shown that 

13 



sub-visual cirrus clouds may cover as much as 70% of the tropics [Wang, 1996] and yet these are 
the clouds that are most difficult to detect using passive sensors and that can even go undetected 
during the daytime by low-pulse-energy lidar systems [Comstock et. al, 2005]. Space-based lidar 
systems such as the current Geosciences Laser Altimetry System (GLAS) [Spinhirne et. al., 2005] 
and CALIPSO [Liu et. al., 2004], scheduled for launch in 2006, have the ability to detect cirrus 
clouds globally and develop statistics of cirrus clouds not possible with passive sensors. However, 
to calculate cirrus cloud optical depths the backscatter measured by space-based lidar must be con- 
verted to extinction assuming some value for the extinction-to-backscatter ratio, otherwise known 
as the lidar ratio. Recent work [Whiteman et. al., 2004] based on earlier cirrus cloud measure- 
ments [Whiteman et. al., 2001a] has shown that this value can vary by a factor of two in very cold 
clouds depending on whether the cloud was hurricane or air-mass-movement induced. Therefore it 
is important to quantify cirrus cloud properties under a range of measurement conditions to assess 
the natural range of variability of the cirrus cloud lidar ratio. 

Generally Raman lidar measurements of cirrus cloud optical depth and extinction-to-backscatter 
ratio have not been made in the daytime. The recent use of pure rotational Raman scattering 
coupled with a Fabry-Perot etalon for temperature profiling has demonstrated the ability to measure 
cirrus cloud extinction during the daytime [Arshinov et. al., 2005]. But cirrus cloud optical depth 
measurements during the daytime using the technologically simpler approach of measuring the 
vibrational q-branch of N 2 have not been demonstrated previously due to poor signal-to-noise 

14 



measurements at cirrus altitudes. Using an interference filter produced under this research with 
a bandwidth of 0.1 nm centered on the Raman vibrational q-branch, Raman lidar measurements 
of cirrus cloud optical depth and extinction-to-backscatter ratio have been made for the first time 
using vibrational Raman scattering during the daytime. Figure 4 shows upward-looking RASL 
measurements of cirrus cloud scattering ratio, optical depth and extinction-to-backscatter ratio 
calculated with 1 -minute temporal resolution. The solar zenith angle was approximately 45 degrees 
during this measurement period. The statistical uncertainty of both the optical depth and lidar ratio 
retrievals is less than 10%. The filter passband was centered on the q-branch of N 2 by tilt-tuning 
of the filter. High resolution spectroscopy indicates that the q-branch of N2 consists of closely 
spaced lines over a spectral interval of approximately 5 cm -1 [Bendtsen and Rasmussen, 2000]. 
This translates to approximately 0.075 nm in wavelength space at the Raman shifted wavelength 
of 386.7 nm. Any variations in laser output wavelength could cause a varying fraction of the q- 
branch intensity to be transmitted by the filter. Therefore, precise control over the laser wavelength 
is desirable for an experimental configuration such as this. The Continuum 9050 laser in use in 
this experiment was not injection-seeded. We observed changes in the transmitted intensity of the 
Raman N 2 signal when the temperature of the laser cooling water changed by ±5C. We concluded 
that these transmitted intensity changes were due to variations in the laser output wavelength. 
Control of the laser cooling water temperature to +/- 1.0C eliminated any noticeable changes in 
the transmitted intensity of Raman scattering from N2 as confirmed by Burleigh pulsed wavemeter 

15 



measurements. This new cirrus cloud measurement capability will permit Raman lidar systems to 

provide useful measurements of cirrus cloud optical quantities during both daytime and nighttime. 
4.3 Carbon Dioxide 



The combination of the use of carbon-based fuels and the reduction in photosynthesis due to the 
clearing of land has caused concentrations of carbon dioxide (CO2) and methane (CH 4 ) to now be 
higher than they have been for at least 100,000 years. The challenge of accurately modeling and 
therefore predicting carbon amounts in the atmosphere is illustrated by the high precision required 
to study some of the key processes driving carbon flux in the atmosphere. Space-based sensors 
are challenged to measure changes in the column content of C0 2 of less than 1%. However, 
most of the short-term variation in the column content of C0 2 is occurring within the atmospheric 
boundary layer where CO2 concentrations may increase by 5 to 10% overnight particularly closest 
to the surface [Bakwin et. al, 1998]. Ground-based and airborne sensors are both closer to the 
region of maximum variation in C0 2 and can be developed more quickly than space-based sensors. 
Therefore, as space-based systems are developed, it makes sense to pursue attractive ground-based 
and airborne technologies that can help improve our understanding of the carbon cycle. Using high 
transmission interference filters fabricated under this research effort, we demonstrate the feasibility 
of using Raman lidar for the simultaneous profiling of water vapor mixing ratio and carbon dioxide 
mixing ratio. 

16 



4.3.1 Numerical simulations of Raman Iidar CO2 mixing ratio measurements 

To study the anticipated signal strength of a Raman lidar measurement of CO2, a numerical model 
that was previously validated for measurements of water vapor mixing ratio [Whiteman et. al., 2001b] 
was used to simulate a ground-based C0 2 Raman lidar system with the RASL specifications of 0.6 
meter telescope, 17.5 Watt laser emitting at 354.7 nm, 0.3 nm filters centered on the 2 v<i CO2 
Raman transition (1285 cm -1 ) and the N 2 Raman vibrational q-branch (2330 cm -1 ). It should be 
noted that the natural quantity that is measured by a Raman lidar, whether in the case of water vapor 
[Whiteman et. al, 1992] or CO2 [Ansmann et. al., 1992b], is the mixing ratio with respect to dry 
air. This is done by using Raman scattering from molecular nitrogen to normalize the water vapor 
or CO2 signal. The numerical model simulates both CO2 and N2 signals based on atmospheric in- 
put profiles and other quantities [Whiteman et. al., 2001b]. The results are shown in figure 5. The 
simulations were performed assuming a 5-hour average. The spatial resolution was. as follows: 
<1.25 km: 75m, 1.25 - 2.0 km: 150m, 2.0 - 2.5 km: 250 m, 2.5 - 3.0 km: 400m, above 3.0 km: 
600m. 

The input to the model included a 10 ppm increase in CO2 at a height of 2.2 km to simulate the 
depletion of CO2 within the mixed layer that occurs during the daytime. Therefore, the input CO2 
profile simulates a possible condition shortly after sunset since these Raman lidar measurements 
can only be made at night due to the weak nature of the Raman C0 2 signal. As shown in figure 
5, the 10 ppm difference between the mixed layer and the free troposphere is easily resolved using 

17 



the measurement parameters that were simulated. The precision of the measurement decreases at 
each change in vertical smoothing such that it remains below 1 PPM at all altitudes up to ~3.5 km 
using the vertical resolutions mentioned. 

On September 19, 2004, RASL was run for 3 hours acquiring what we believe to be the first 
simultaneous remote profile measurements of atmospheric CO2 and H 2 mixing ratio. These are 
likely the first ground-based CO2 profile measurements extending into the free troposphere as well. 
The CO2 measurements were calibrated based on ground-based measurements of CO2 acquired at 
the same time. The C0 2 calibration obtained must therefore be considered only approximate. 
The water vapor measurements were calibrated by forcing the total precipitable water of the lidar 
profile to equal that measured by a collocated GPS sensor. Both the C0 2 and H2O have been 
analyzed such that the vertical resolution is 300 m between 1 and 2 km, 400 m between 2 and 3 
km, 500 m between 3 and 4 km, and 600 m above 4 km. The precision of the CO2 mixing ratio 
measurement obtained with these resolutions, determined from the signal strength of the CO2 and 
N2 data assuming Poisson statistics, remains below 1.5 PPM for altitudes less than 4 km. The 
precision of the C0 2 measurement is generally consistent with the model predictions shown in 
figure 5. The standard error bars plotted on the water vapor mixing ratio data shown in figure 5 are 
imperceptible on this scale. 

Error sources in the measurement of C0 2 using Raman lidar The only known previous 

18 



measurements of atmospheric CO2 (2 v 2 : 1285 cm -1 ) using Raman lidar were made by a Ph.D. 
student [Riebesell, 1990] [?] working at the GKSS Institute in Hamburg, Germany in the late 
1980s. The conclusions based on that research were that useful CO2 measurements by Raman 
lidar were unlikely because the interference from rotational lines of O2 was difficult to determine 
and fluorescence of either optics or atmospheric particles could contaminate the measurement at 
the ~lppm level. However, this earlier research was conducted using a XeCl excimer laser, which 
has an output spectrum that spans approximately 0.4 ran. This broad spectrum makes the sepa- 
ration of O2 and C0 2 more difficult than the present use of narrow-band interference filters and 
a Nd:YAG laser with spectral output of -0.02 ran. Calculations indicate that the contribution of 
O2 rotational lines to the measured CO2 signal in the present configuration is approximately 1% 
(-3-4PPM). Rotational line strength modeling [Whiteman et. al., 2001b] can be used to account 
for this contribution reducing the uncertainty in the C0 2 measurement due to O2 rotational line 
interference to much less than 1 PPM. 

A careful study of fluorescence of both optical components and atmospheric aerosols would 
be required as a part of further developing and validating a Raman lidar CO2 profiling system. 
Preliminary measurements acquired using a scanning spectrometer coupled to a Raman lidar re- 
ceiver indicated no significant fluorescence contribution in the CO2 spectral region, even though 
fluorescence due to aerosols was observed at longer wavelengths during the same measurement 
period. 

19 



5 Summary and Conclusions 

Research conducted under the NASA Advanced Component Technology (ACT) Program has re- 
sulted in the construction of narrow-band interference filters with approximately twice the trans- 
mission of previous technology. This was accomplished while maintaining the same of out-of-band 
blocking that is required for Raman lidar application. A technique of interference filter construc- 
tion based on ion assisted deposition (IAD) was found to produce filters with the best overall 
performance. Considerable effort was expended addressing four separate concerns in filter fab- 
rication: improving spatial uniformity, minimizing scattering losses, minimizing reflection losses 
and minimizing absorption losses. At the end of the effort, interference filters possessing band- 
widths as narrow as 0.1 nm and peak transmissions of twice what was previously possible were 
produced. However, the yield of spatially uniform filters was somewhat low indicating that more 
research in process control is required in order to permit easier manufacture of such narrow filters. 

Improved Raman lidar measurements were demonstrated using filters produced under this re- 
search that were designed to measure Raman scattering from water vapor, nitrogen and carbon 
dioxide. The measurements were acquired using the Raman Airborne Spectroscopic Lidar (RASL) 
operating from the ground in an upward-looking configuration. Water vapor measurements pos- 
sessing 2-minute temporal and 60 - 210 m spatial resolution were presented. Except for a near- 
range zone of approximately 600 meters, where random errors increase due to the dynamic range 

20 



suppression that is inherent in the narrow field-of-view design of RASL, random errors remained 
below 4% up to an altitude of approximately 4km. This measurement capability is shown to be 
sufficient to quantify boundary layer turbulence under daytime conditions. 

For the first time, cirrus cloud optical depth and extfnction-to-backscatter ratio (lidar ratio) were 
quantified in the daytime using a measurement of Raman vibrational scattering from molecular 
nitrogen. Using 1 -minute temporal resolution, both optical depth and lidar ratio were quantified 
with approximately 10% uncertainty under daytime conditions where the solar zenith angle was 
approximately 45-50 degrees. This new measurement capability will permit cirrus cloud statistics 
to be acquired throughout the diurnal cycle using Raman lidar. 

The final measurements that were shown are what we believe to be the first simultaneous re- 
mote measurements of atmospheric carbon dioxide and water vapor mixing ratio extending into 
the free troposphere. The carbon dioxide measurements were approximately calibrated based on a 
ground-based measurement of CC>2- The random error of the measurements agreed well with pre- 
dictions based on numerical simulation. Error sources in the measurement of CO2 using Raman 
lidar were considered. The interference of rotational lines from 2 were estimated to be small. 
Aerosol fluorescence was studied briefly and found to not contribute signal in the spectral band of 
CO2. Additional error sources such as the lidar system overlap function and the differential trans- 
mission of the atmosphere must also be studied. When all of these error sources are considered, 
it is unlikely that a Raman lidar measurement of CO2 with absolute accuracy of 1 PPM could be 

21 



achieved. However, for the study of carbon sources and sinks a quantification of nocturnal changes 
in C0 2 with a precision of 1 PPM (as opposed to an absolute accuracy of a 1 PPM) is sufficient 
to provide useful information for modeling efforts that are designed to improve our understanding 
of carbon processes in the atmosphere. Therefore, the results presented here indicate that contin- 
ued development of Raman lidar profiling of CO2 is a worthwhile research effort since the signal 
strength exists for highly precise nocturnal C0 2 profile measurements to be made and for their 
correlation with H2O to be studied. 

6 Acknowledgements 

The authors wish to acknowledge the support of the NASA Advanced Component Technology 
(ACT) program and Instrument Incubator Program (IIP) for support of these efforts. The ra- 
diosonde data were obtained from the Howard University Beltsville Research Campus under a 
program supported by the Maryland Department of the Environment. 

7 References 

[Ansmannet. al., 1990] AAnsmann, M. RiebeselL and C. Weitkamp, 1990: "Measurement of 
atmospheric aerosol extinction profiles with a Raman lidar," Optics Letters, 15, 13, 746-748. 

[Ansmann et. al, 1992a] Ansmann, A., U. Wandinger, M. Riebesell, C. Weitkamp, W. Michaelis, 
1992: Independent measurement of extinction and backscatter profiles in cirrus clouds by 

22 



using a combined Raman elastic-backscatter Mar, Appl. Opt., 31, No. 33, 7113-7131. 

[Ansmann et. al., 1992b] A. Ansmann, M. Riebesell, C. Weitkamp, E. Voss. W. Lahmann, W. 
Michaelis,1992: "Combined Raman Elastic-backscatter lidar for vertical profiling of moisture, 
aerosol extinction, backscatter, and lidar ratio, Appl. Phys. B., 55 (1), 18-28. 

[Arshinov et. al.,2005] Arsbinov, Y. S. Bobrovnikov, I. Serikov, A. Ansmann, U. Wandinger, 
D. Althausen, I. Mattis, D. Miiller, 2005: "Daytime operation of a pure rotational Raman lidar 
by use of a Fabry-Perot interferometer", Appl. Opt, 44 (17): 3593-3603 (2005). 

[Bakwinet. al, 1998] Bakwin, P.S., P. P. Tans, D. F. Hurst, C. Zhao, 1998" "Measurements 
of carbon dioxide on very tall towers: results of the NOAA/CMDL program", Tellus, 50B, 
401-415. 

[Behrendt et. al., 2002] Behrendt A, T. Nakamura, M. Onishi, R. Baumgart, T. Tsuda, 2002: 
"Combined Raman lidar for the measurement of atmospheric temperature, water vapor, parti- 
cle extinction coefficient, and particle backscatter coefficient", Appl. Opt., 41 (36): 7657-7666. 

[Bendtsen and Rasmussen, 2000] Bendtsen, J. and F. Rasmussen, 2000: "High- 

resolution incoherent Fourier transform Raman spectrum of the fundamental band of 14 N2" , J. 
Raman Spectrosc. 31, 433-438. 



23 



[Bosenberg, 2005] Bosenberg, J., "Differential- Absorption Lidar for Water Vapor and Tem- 

perature Profiling", in Lidar Range-resolved Optical Remote Sensing C. Weitkamp, ed. (Springer 
Series in Optical Sciences 102, 2005) 

[Comstock et. al, 2005] Comstock, J. M., T.P. Ackerman, and G. G. Mace, 2002: "Ground based 
lidar and radar remote sensing of tropical cirrus clouds at Nauru Island: Cloud statistics and 
radiative impacts", J. Geophys. Res., 107, 4714. 

[Demoz et. al., 2005] Demoz BB, Starr DOC, Evans KD, Lare AR, Whiteman DN, Schwem- 

mer G, Ferrare RA, Goldsmith JEM, Bisson SE, 2005: "The cold front of 15 April 1994 over 
the central United States. Part I: Observations", Mon. Wea. Rev. 133 (6): 1525-1543. 

[Demoz et. al., 2006] Demoz, B. C. Flamant, T. Weckwerth, D. Whiteman, K. Evans, F. Fabry, 

P Di Girolamo, D. Miller, B. Geerts, W. Brown, G. Schwemmer, B. Gentry, W. Feltz, and Z. 
Wang, 2006: "The Dryline on 22 May 2002 during IHOP-2002: Convective-Scale Measure- 
ments at the Profiling Site", Monthly Weather Review, Vol. 134, No. 1, 294-310. 

[Di Girolamo et. al, 2004] Di Girolamo, P., R. Marchese, D. N. Whiteman, B. B. Demoz, 2004: 
Rotational Raman Lidar measurements of atmospheric temperature in the UV, Geophys. Res. 
Lett, 31, L01106, doi:10.1029/2003GL018342. 

[Ferrare et. al., 2005] Ferrare, R. A., D.D. Turner, M. Clayton, B. Schmid, J. Redemann, D. 

24 



Covert, R. Elleman, J. Ogren, E. Andrews, J.E.M. Goldsmith, and H. Johsson, 2005: Raman 
lidar measurements of aerosols and water vapor over the Southern Great Plains during the May 
2003 Aerosol IOR J. Geophys. Res., submitted. 

[Liu et. al, 2004] Liu, Z., M. Vaughan, D. Winker, C. Hostetler, L. Poole, D. Hlavka, W. 

Hart, M. McGill, 2004: "Use of probability distribution functions for discriminating between 
cloud and aerosol in lidar backscatter data", J. Geophys. Res., Vol. 109,D15202, doi: 10.1 029/2004 JD0047 

[MacLeod et. al, 1989] H.A. MacLeod, H. A., Thin Film Optical Filters, 2nd edition McGraw- 
Hill Publishing Co., New York 1989 pp. 412-420. 

[Mehiet. al, 1989] Melfi, S. H., D. N. Whiteman, R. A. Ferrare, 1989: "Observation of 

atmospheric fronts using Raman lidar moisture measurements", J. App. Meteor., 28, 789-806. 

[Pawlewicz, 1998] " Pawlewicz, W. T., "Property and processing comparison of optical coat- 
ings made by ion assisted evaporation and magnetron sputtering" Proceedings of Materials 
Research Society Fall Meeting, Boston, 1998. 

[Potter and Simons, 1993] Potter, J. R., and J.C. Simons, "Stability of refractory oxide narrowband 
interference filters" SPIE, vol. 1952, pp. 186-191, April 1993 

[Reichardt et. al, 2002] Reichardt, J., S. Reichardt, A. Behrendt, and T. J. McGee, Correlations 



25 



among the optical properties of cirrus-cloud particles: Implications for spaceborne remote 
sensing, Geophys. Res. Lett., 29, 1029-1032, 2002. 

[Riebesell, 1990] Riebesell, M., 1990: "Raman-LidarzurFernmessung von Wasserdampf- 

und Kohlendioxid-Hohenprofilen in der Tropsphare", PhD Thesis, GKSS document 90/E/13, 
127 pp. 

[Schinket. al., 1995] Schink, H., E. Masetti, M. Montecchi, D. Flori, "Optical absorption 

of Hf0 2 and mixed Hf02/Al 2 03 thin films in the 230-700 nm range" Optical Interference 
Coatings, OSA vol. 17, pp.307-308, June, 1995. 

[Spinhirne et. al, 2005] Spinhirne JD, Palm SP, Hart WD, Hlavka DL, Welton EJ, 2005 : "Cloud 
and aerosol measurements from GLAS: Overview and initial results", Geophys. Res. Lettr., 
32(22): Art. No. L22S03. 

[Turner et. al., 2000] Turner, D. D., W. F. Feltz, R. A. Ferrare, 2000: "Continuous Water Pro- 

files from Operational Ground-based Active and Passive Remote Sensors", Bull. Amer. Soc, 
81,1301-1317. 

[Wang, 1996] Wang, P.-H., P. Minnis, M. P. McCormick, G. S. Kent, and K. M. 

Skeens, 1996: "A 6-year climatology of cloud occurrence frequency from Stratospheric Aerosol 
and Gas Experiment II observations (1985-1990)", J. Geophys. Res., 101, 29,407-29,429. 

26 



[Weckwerthet.al.,2004] Weckwerth T. M., D. B. Parsons, S. E. Koch, J. A. Moore, M. A. 
LeMone, B. B. Demoz, C. Flamant, B. Geerts, J. H.Wang, W. F. Feltz, 2004: "An overview 
of the International H20 Project (IHOP_2002) and some preliminary highlights", Bull. Amer. 
Meteor. Soc, 85 (2) 253. 

[Whiteman et. al, 1992] Whiteman, D.N., S.H. Mem, and R.A. Ferrare, 1992: Raman lidar sys- 
tem for the measurement of water vapor and aerosols in the earth's atmosphere", Appl. Opt., 
31, No. 16,3068-3082. 

[Whiteman et. al., 2001a] Whiteman, D.N., K. D. Evans, B. Demoz, D. O'C. Starr, E. Eloranta, 
D. Tobin, W Feltz, G. J. Jedlovec, S. I. Gutman, G. K. Schwemmer, M. Cadirola, S. H. Melfi, 
F. J. Schmidlin, 2001 : Raman lidar measurements of water vapor and cirrus clouds during the 
passage of hurricane Bonnie, J. ofGeophys. Res., 106, No. D6, 521 1-5225. 

[Whiteman et. al, 2001b] Whiteman, D.N., G. Schwemmer, T. Berkoff, H. Plotkin, L. Ramos- 
Izquierdo, G. Pappalardo, 2001: Performance modeling of an airborne Raman water vapor 
lidar, Appl. Opt.,40, No. 3, 375-390. 

[Whiteman et. al, 2004] Whiteman, D.N., B. Demoz, Z. Wang: 2004, "Subtropical cirrus cloud 
extinction to backscatter ratios measured by Raman Lidar during CAMEX-3", Geophys. Res. 
Lett, 31, L12105, doi:10.1029/2004GL020003. 



27 



[Whiteman et. al., 2006a] Whiteman, D.N., B. Demoz, P. Di Girolamo, J. Comer, I. Veselovskii, 
K. Evans, Z. Wang, M. Cadirola, K. Rush, D. Sabatino, G. Schwemmer, B. Gentry, S. H. Melfi, 
B. Mielke, D. Venable, T. Van Hove, E. Browell, R. Ferrare, S. Ismail, J. Wang, 2006: Raman 
Water Vapor Lidar Measurements During the International H2O Project. I. Instrumentation and 
Analysis Techniques,,/ Atmos. Oceanic Technol, 23, 157-169. 

[Whiteman et. al, 2006b] Whiteman, D.N., B. Demoz, P. Di Girolamo, J. Comer, I. Veselovskii, 
K. Evans, Z. Wang, M. Cadirola, K. Rush, D. Sabatino, G. Schwemmer, B. Gentry, S. H. 
Melfi, B. Mielke, D. Venable, T. Van Hove, E. Browell, R. Ferrare, S. Ismail, J. Wang, 2006: 
Raman Water Vapor Lidar Measurements During the International H2O Project, n. Instrument 
Comparisons and Case Studies, J. Atmos. Oceanic Technol, 23, 157-169. 

[Wulfmeyer et. al., 2006] Wulfmeyer, V. H-S Bauer, M. Grzeschik, A. Behrendt, F. Vandenberghe, 
E. V Browell, S. Ismail and R. A. Ferrare, 2006: "Four-Dimensional Variational Assimilation 
of Water Vapor Differential Absorption Lidar Data: The First Case Study within IHOP_2002", 
Mon. Wea. Rev., 134 (1), 209-230. 



28 



July 26, 2005 Water Vapor Mixing Ratio 




(g/kg) 



I 



20 



18 20 

Time (UT) 



Figure 1 : Water vapor mixing ratio measurements made by the upward looking RASL instrument 
using the narrow band water vapor and nitrogen interference niters developed under this research 
effort. 



8 Figures 



29 




6 3 10 

Mixing Ratio (g/Kg) 



Figure 2: A comparison of RASL measurements of water vapor mixing ratio and those of a ra- 
diosonde launched approximately 10 km away. The layering of the features is very similar between 
these two sites during these early morning measurements. 



30 




5 10 15 

Mixing Ratio (g/Kg) 



20 



20 40 60 80 

Mixing Ratio Error (%) 



100 



Figure 3: RASL profiles of water vapor mixing ratio and the random error in water vapor mixing 
ratio measured at 13, 18 and 26.5 UT. 



31 




18.4 



15.S 



15.8 15.7 

Tiros <UT} 



15,8 




1S.fi 1S.7 

TimeOJT) 



Figure 4: Measurements of cirrus cloud scattering ratio, optical depth and extinction-to-backscatter 
(lidar) ratio made during the daytime with a solar zenith angle of" -45-50 degrees using 1 minute 
temporal resolution. 



32 






C0 2 Input 



Simulation 



300 320 340 360 

Number Density (ppm) 



380 




0.5 1 1.5 2 2.5 

Number Density (ppm) 



Figure 5 : Model simulations of ground-based profiling of CO2 during the nighttime. The parame- 
ters simulated are 0.6 m telescope and 17. 5W UV laser with an averaging time of 5 hours. The 
resultant precision is below lppm for all altitudes below 3.5 km with vertical resolution ranging 
from 75 m to 600 m. A free tropospheric transition was simulated at approximately 2 km where 
the vertical resolution of the simulation was 250 m. 



33 



I 3 

"a 

3 





': -K-- . 




W" 




k:.\ 




r—- 1 



350 355 360 365 370 
Est. C0 2 Mixing Ratio (PPM) 




0.6 0.8 1 1.2 1.4 1.6 
Precision (PPM) 



1.8 



2 4 6 8 10 12 14 
H 2 Mixing Ratio (g/kg) 



Figure 6: The carbon dioxide mixing ratio is shown on the left using an approximate calibration 
derived from ground-based measurements. The precision of the CO2 measurent is shown in the 
second panel. The simultaneously acquired water vapor mixing ratio measurement is also shown. 
The averaging time for these measurements was 3 hours. 



34 



CWL BW T (%) General Additional block- Measurement 

(+0.02/0.00 (±0.02) blocking ing 

nm) 

1) 407.5 0.25 70 OD6 @ OD12 @354.7nm Raman water va- 

200-1200 OD8 @ 375- por 
nm 387nm OD9 

@532&1064nm 



2) 386.68 


A 1 
U.l 


£A 

uu 


OD6 @ 

200-1200 

nm 


OD12 

@354.7nm OD9 
@532&1064nm 


i\.aman niuOgen 


3)371.71 


0.1 


40 


OD6 @ 


OD12 @354.7nm 


Raman carbon 








200-1200 


OD7 @375- 


dioxide 








nm 


387nm 





Table 1 : Specifications of interference filters discussed in this article. BW refers to the full width 
half maximum bandwidth of the filter, CWL to the center wavelength of the filter, T to the trans- 
mission. 



Material 


Surface Roughness (nm RMS) 


T (%) 


Polished B-270 


0.15 


43 


Polished UV fused silica 


0.12 


56 


Float Soda Lime Glass 


0.048 


73 



Table 2: Substrate surface roughness versus transmission. 



9 Tables 



35